📄 doublemetaphone.java
字号:
//-- Wasserman should match Vasserman --//
result.append('A', 'F');
} else {
//-- need Uomo to match Womo --//
result.append('A');
}
index++;
} else if ((index == value.length() - 1 && isVowel(charAt(value, index - 1))) ||
contains(value, index - 1,
5, "EWSKI", "EWSKY", "OWSKI", "OWSKY") ||
contains(value, 0, 3, "SCH")) {
//-- Arnow should match Arnoff --//
result.appendAlternate('F');
index++;
} else if (contains(value, index, 4, "WICZ", "WITZ")) {
//-- Polish e.g. "filipowicz" --//
result.append("TS", "FX");
index += 4;
} else {
index++;
}
}
return index;
}
/**
* Handles 'X' cases
*/
private int handleX(String value,
DoubleMetaphoneResult result,
int index) {
if (index == 0) {
result.append('S');
index++;
} else {
if (!((index == value.length() - 1) &&
(contains(value, index - 3, 3, "IAU", "EAU") ||
contains(value, index - 2, 2, "AU", "OU")))) {
//-- French e.g. breaux --//
result.append("KS");
}
index = contains(value, index + 1, 1, "C", "X") ? index + 2 : index + 1;
}
return index;
}
/**
* Handles 'Z' cases
*/
private int handleZ(String value, DoubleMetaphoneResult result, int index,
boolean slavoGermanic) {
if (charAt(value, index + 1) == 'H') {
//-- Chinese pinyin e.g. "zhao" or Angelina "Zhang" --//
result.append('J');
index += 2;
} else {
if (contains(value, index + 1, 2, "ZO", "ZI", "ZA") || (slavoGermanic && (index > 0 && charAt(value, index - 1) != 'T'))) {
result.append("S", "TS");
} else {
result.append('S');
}
index = charAt(value, index + 1) == 'Z' ? index + 2 : index + 1;
}
return index;
}
//-- BEGIN CONDITIONS --//
/**
* Complex condition 0 for 'C'
*/
private boolean conditionC0(String value, int index) {
if (contains(value, index, 4, "CHIA")) {
return true;
} else if (index <= 1) {
return false;
} else if (isVowel(charAt(value, index - 2))) {
return false;
} else if (!contains(value, index - 1, 3, "ACH")) {
return false;
} else {
char c = charAt(value, index + 2);
return (c != 'I' && c != 'E')
|| contains(value, index - 2, 6, "BACHER", "MACHER");
}
}
/**
* Complex condition 0 for 'CH'
*/
private boolean conditionCH0(String value, int index) {
if (index != 0) {
return false;
} else if (!contains(value, index + 1, 5, "HARAC", "HARIS") &&
!contains(value, index + 1, 3, "HOR", "HYM", "HIA", "HEM")) {
return false;
} else if (contains(value, 0, 5, "CHORE")) {
return false;
} else {
return true;
}
}
/**
* Complex condition 1 for 'CH'
*/
private boolean conditionCH1(String value, int index) {
return ((contains(value, 0, 4, "VAN ", "VON ") || contains(value, 0,
3, "SCH")) ||
contains(value, index - 2, 6, "ORCHES", "ARCHIT", "ORCHID") ||
contains(value, index + 2, 1, "T", "S") ||
((contains(value, index - 1, 1, "A", "O", "U", "E") || index == 0) &&
(contains(value, index + 2, 1, L_R_N_M_B_H_F_V_W_SPACE) || index + 1 == value.length() - 1)));
}
/**
* Complex condition 0 for 'L'
*/
private boolean conditionL0(String value, int index) {
if (index == value.length() - 3 &&
contains(value, index - 1, 4, "ILLO", "ILLA", "ALLE")) {
return true;
} else if ((contains(value, index - 1, 2, "AS", "OS") ||
contains(value, value.length() - 1, 1, "A", "O")) &&
contains(value, index - 1, 4, "ALLE")) {
return true;
} else {
return false;
}
}
/**
* Complex condition 0 for 'M'
*/
private boolean conditionM0(String value, int index) {
if (charAt(value, index + 1) == 'M') {
return true;
}
return contains(value, index - 1, 3, "UMB")
&& ((index + 1) == value.length() - 1 || contains(value,
index + 2, 2, "ER"));
}
//-- BEGIN HELPER FUNCTIONS --//
/**
* Determines whether or not a value is of slavo-germanic orgin. A value is
* of slavo-germanic origin if it contians any of 'W', 'K', 'CZ', or 'WITZ'.
*/
private boolean isSlavoGermanic(String value) {
return value.indexOf('W') > -1 || value.indexOf('K') > -1 ||
value.indexOf("CZ") > -1 || value.indexOf("WITZ") > -1;
}
/**
* Determines whether or not a character is a vowel or not
*/
private boolean isVowel(char ch) {
return VOWELS.indexOf(ch) != -1;
}
/**
* Determines whether or not the value starts with a silent letter. It will
* return <code>true</code> if the value starts with any of 'GN', 'KN',
* 'PN', 'WR' or 'PS'.
*/
private boolean isSilentStart(String value) {
boolean result = false;
for (int i = 0; i < SILENT_START.length; i++) {
if (value.startsWith(SILENT_START[i])) {
result = true;
break;
}
}
return result;
}
/**
* Cleans the input
*/
private String cleanInput(String input) {
if (input == null) {
return null;
}
input = input.trim();
if (input.length() == 0) {
return null;
}
return input.toUpperCase();
}
/**
* Gets the character at index <code>index</code> if available, otherwise
* it returns <code>Character.MIN_VALUE</code> so that there is some sort
* of a default
*/
protected char charAt(String value, int index) {
if (index < 0 || index >= value.length()) {
return Character.MIN_VALUE;
}
return value.charAt(index);
}
/**
* Shortcut method with 1 criteria
*/
private static boolean contains(String value, int start, int length,
String criteria) {
return contains(value, start, length,
new String[] { criteria });
}
/**
* Shortcut method with 2 criteria
*/
private static boolean contains(String value, int start, int length,
String criteria1, String criteria2) {
return contains(value, start, length,
new String[] { criteria1, criteria2 });
}
/**
* Shortcut method with 3 criteria
*/
private static boolean contains(String value, int start, int length,
String criteria1, String criteria2,
String criteria3) {
return contains(value, start, length,
new String[] { criteria1, criteria2, criteria3 });
}
/**
* Shortcut method with 4 criteria
*/
private static boolean contains(String value, int start, int length,
String criteria1, String criteria2,
String criteria3, String criteria4) {
return contains(value, start, length,
new String[] { criteria1, criteria2, criteria3,
criteria4 });
}
/**
* Shortcut method with 5 criteria
*/
private static boolean contains(String value, int start, int length,
String criteria1, String criteria2,
String criteria3, String criteria4,
String criteria5) {
return contains(value, start, length,
new String[] { criteria1, criteria2, criteria3,
criteria4, criteria5 });
}
/**
* Shortcut method with 6 criteria
*/
private static boolean contains(String value, int start, int length,
String criteria1, String criteria2,
String criteria3, String criteria4,
String criteria5, String criteria6) {
return contains(value, start, length,
new String[] { criteria1, criteria2, criteria3,
criteria4, criteria5, criteria6 });
}
/**
* Determines whether <code>value</code> contains any of the criteria
starting
* at index <code>start</code> and matching up to length <code>length</code>
*/
protected static boolean contains(String value, int start, int length,
String[] criteria) {
boolean result = false;
if (start >= 0 && start + length <= value.length()) {
String target = value.substring(start, start + length);
for (int i = 0; i < criteria.length; i++) {
if (target.equals(criteria[i])) {
result = true;
break;
}
}
}
return result;
}
//-- BEGIN INNER CLASSES --//
/**
* Inner class for storing results, since there is the optional alternate
* encoding.
*/
public class DoubleMetaphoneResult {
private StringBuffer primary = new StringBuffer(getMaxCodeLen());
private StringBuffer alternate = new StringBuffer(getMaxCodeLen());
private int maxLength;
public DoubleMetaphoneResult(int maxLength) {
this.maxLength = maxLength;
}
public void append(char value) {
appendPrimary(value);
appendAlternate(value);
}
public void append(char primary, char alternate) {
appendPrimary(primary);
appendAlternate(alternate);
}
public void appendPrimary(char value) {
if (this.primary.length() < this.maxLength) {
this.primary.append(value);
}
}
public void appendAlternate(char value) {
if (this.alternate.length() < this.maxLength) {
this.alternate.append(value);
}
}
public void append(String value) {
appendPrimary(value);
appendAlternate(value);
}
public void append(String primary, String alternate) {
appendPrimary(primary);
appendAlternate(alternate);
}
public void appendPrimary(String value) {
int addChars = this.maxLength - this.primary.length();
if (value.length() <= addChars) {
this.primary.append(value);
} else {
this.primary.append(value.substring(0, addChars));
}
}
public void appendAlternate(String value) {
int addChars = this.maxLength - this.alternate.length();
if (value.length() <= addChars) {
this.alternate.append(value);
} else {
this.alternate.append(value.substring(0, addChars));
}
}
public String getPrimary() {
return this.primary.toString();
}
public String getAlternate() {
return this.alternate.toString();
}
public boolean isComplete() {
return this.primary.length() >= this.maxLength &&
this.alternate.length() >= this.maxLength;
}
}
}
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -